Nucleosides and their analogues represent an important class of chemotherapeutic agents with antiviral, anticancer, immunomodulatory and antibiotic activities. Nucleoside analogues such as 3'-azido-3'-deoxythymidine (AZT), 2',3'-dideoxyinosine (ddI), 2',3'-dideoxycytidine (ddC), 3'-deoxy-2',3'-didehydrothymidine (d.sub.4 T) and (-)-2'-deoxy-3'-thiacytidine (3TC.TM.) are clinically approved for the treatment of infections caused by the human immunodeficiency viruses. 2'-Deoxy-2'-methylidenecytidine (DMDC, Yamagami et al. Cancer Research 1991, 51, 2319) and 2'-deoxy-2',2'-difluorocytidine (gemcytidine, Hertel et al. J. Org. Chem. 1988, 53, 2406) are nucleoside analogues with antitumor activity. A number of C-8 substituted guanosines such as 7-thia-8-oxoguanosine (Smee et al. J. Biol. Response Mod. 1990, 9, 24) 8-bromoguanosine and 8-mercaptoguanosine (Wicker et al. Cell Immunol. 1987, 106, 318) stimulate the immune system and induce the production of interferon. All of the above biologically active nucleosides are single enantiomers.
Recently, several members of the 3'-heterosubstituted class of 2',3'-dideoxynucleoside analogues such as 3TC.TM.(Coates et al. Antimicrob. Agents Chemother. 1992, 36, 202), (-)-FTC (Chang et al. J. Bio. Chem. 1992, 267, 13938-13942) (-)-dioxolane C (Kim et al. Tetrahedron Lett. 1992, 33, 6899) have been reported to possess potent activity against HIV and HBV replication and possess the .beta.-L absolute configuration. (-)-Dioxolane C has been reported to possess antitumor activity (Grove et al. Cancer Res. 1995, 55, 3008-3011). The dideoxynucleoside analogues (-)-dOTC and (-)-dOTFC (Mansour et al. J. Med. Chem. 1995, 38, 1-4) were selective in activity against HIV-1.
For a stereoselective synthesis of nucleoside analogues, it is essential that the nucleobase be introduced predominately with the desired relative stereochemistry without causing anomerization in the carbohydrate portion. One approach to achieve this is to modify the carbohydrate portion of a preassembled nucleoside by a variety of deoxygenation reactions (Chu et al. J. Org. Chem. 1989, 54, 2217-2225; Marcuccio et al. Nucleosides Nucleotides 1992, 11, 1695-1701; Starrett et al. Nucleosides Nucleotides 1990, 9, 885-897, Bhat et al. Nucleosides Nucleotides 1990, 9, 1061-1065). This approach however is limited to the synthesis of those analogues whose absolute configuration resembles that of the starting nucleoside and would not be practical if lengthy procedures are required to prepare the starting nucleoside prior to deoxygenation as would be the case for .beta.-L dideoxynucleosides. An alternative approach to achieve stereoselectivity has been reported which requires assembling the nucleoside analogue by a reaction of a base or its synthetic precursor with the carbohydrate portion under Lewis acid coupling procedures or SN-2 like conditions.
It is well known in the art that glycosylation of bases to dideoxysugars proceed in low stereoselectivity in the absence of a 2'-substituent on the carbohydrate rings capable of neighboring group participation. Okabe et al. (J. Org. Chem. 1988, 53, 4780-4786) reported the highest ratio of .beta.:.alpha. isomers of ddC of 60:40 with ethylaluminium dichloride as the Lewis acid. However, with a phenylselenenyl substituent at the C-2 position of the carbohydrate (Chu et al. J. Org. Chem. 1980, 55, 1418-1420; Beach et al. J. Org. Chem. 1992, 57, 3887-3894) or a phenylsulfenyl moiety (Wilson et al. Tetrahedron Lett. 1990, 31, 1815-1818) the .beta.:.alpha. ratio increases to 99:1. To overcome problems of introducing such substituents with the desired .alpha.-stereochemistry, Kawakami et al. (Nucleosides Nucleotides 1992, 11, 1673-1682) reported that disubstitution at C-2 of the sugar ring as in 2,2-diphenylthio-2,3-dideoxyribose affords nucleosides in the ratio of .beta.:.alpha.=80:20 when reacted with silylated bases in the presence of trimethylsilyltriflate (TMSOTf) as a catalyst. Although this strategy enabled the synthesis of the .beta.-anomer, removal of the phenylthio group proved to be problematic.
Due to the limited generality in introducing the C-2 substituent stereoselectively, synthetic methodologies based on electrophilic addition of phenyl sulfenyl halides or N-iodosuccinimides and nucleobases to furanoid glycal intermediates have been reported (Kim et al. Tetrahedron Lett. 1992, 33, 5733-5376; Kawakami et al. Heterocycles 1993, 36, 665-669; Wang et al. Tetrahedron Lett. 1993, 34, 4881-4884; El-laghdach et al. Tetrahedron Lett. 1993, 34, 2821-2822). In this approach, the 2'-substituent is introduced in situ however, multistep procedures are needed for removal of such substituents.
SN-2 like coupling procedures of 1-chloro and 1-bromo 2,3-dideoxysugars have been investigated (Farina et al. Tetrahedron Lett. 1988, 29, 1239-1242; Kawakami et al. Heterocycles 1990, 31, 2041-2053). However, the highest ratio of .beta. to .alpha. nucleosides reported is 70:30 respectively.
In situ complexation of metal salts such as SnCl.sub.4 or Ti(O-Pr).sub.2 Cl.sub.2 to the .alpha.-face of the sugar precursor when the sugar portion is an oxathiolanyl or dioxolanyl derivative produces .beta.-pyrimidine nucleosides (Choi et al. J. Am. Chem. Soc. 1991, 113, 9377-9379). Despite the high ratio of .beta.- to .alpha.-anomers obtained in this approach, a serious limitation with enantiomerically pure sugar precursor is reported leading to racemic nucleosides (Beach et al. J. Org. Chem. 1992, 57, 2217-2219; Humber et al. Tetrahedron Lett. 1992, 32, 4625-4628; Hoong et al. J. Org. Chem. 1992, 57, 5563-5565). In order to produce one enantiomeric form of racemic nucleosides, enzymatic and chemical resolution methods are needed. If successful, such methods would suffer from a practical disadvantage of wasting half of the prepared material.
As demonstrated in the above examples, the art lacks an efficient method to generate .beta.-nucleosides. In particular, with sugar precursors carrying a protected hydroxymethyl group at C-4', low selectivity is encountered during synthesis of .beta.-isomers or racemization problems occur. Specifically, the art lacks a method of producing stereoselectively dioxolanes from sugar intermediates carrying a C-2 protected hydroxymethyl moiety without racemization. Therefore, a general stereoselective synthesis of biologically active .beta.-nucleoside analogues is an important goal.
International patent application publication no. WO92/20669 discloses a method of producing dioxolanes stereoselectively by coupling sugar intermediates carrying C-2 ester moieties with silylated nucleobases and subsequently reducing the C-2 ester group to the desired hydroxymethyl group. However, over reduction problems in the pyrimidine base have been disclosed (Tse et al. Tetrahedron Lett. 1995, 36, 7807-7810).
Nucleoside analogues containing 1,3-dioxolanyl sugars as mimetics of 2',3'-dideoxyfuranosyl rings have been prepared by glycosylating silylated purine and pyrimidine bases with 1,3-dioxolanes containing a C-2 hydroxymethyl and C-4 acetoxy substituents. The crucial coupling reaction is mediated by trimethylsilytriflate (TMSOT.sup.f) or iodotrimethylsilane (TMSI) and produces a mixture of .beta. and .alpha.-anomers in 1:1 ratio (Kim et al. J. Med. Chem. 1992, 35, 1987-1995 and J. Med. Chem. 1993, 36, 30-37; Belleau et al. Tetrahedron Lett. 1992, 33, 6948-6952; and Evans et al. Tetrahedron Asymmetry 1992, 4, 2319-2322). By using metal salts as catalysts the .beta.-nucleoside is favoured (Choi et al. J. Am. Chem. Soc. 1991, 113, 9377-9379) but racemization or loss of selectivity become a serious limitation (Jin et al. Tetrahedron Asymmetry 1993, 4, 2111-2114).